Feature level fusion using physical biometric traits

Abstract

Biometric is a measuring technique that is utilised to discern a person's identity through a physiological or behavioural feature. Each modality on its own cannot always be reliable for recognition. The multimodal biometric system offers several advantages over the traditional biometric system. In order to overcome the limitations of the unimodal biometrics, the fusion of multimodal biometric traits have been used. In this paper, a multimodal biometric based on fingerprint, palm print and knuckle print is presented. Gabor filter is employed to extract features from these biometric traits. The features extracted from these biometric traits are normalised and fused together. The features are fused by feature level fusion, and the result of the proposed method provides efficient authentication, by minimising the FAR by 1.43%, and the framework achieves satisfactory performance.